Ok.
Where to start? Let's start by the beginning, which was 1 year ago when i started that thread:
World's best midrange Blind Testing - Need your help.
Now, 100+ pages and 65,000 views later i've come to some conclusions that are... uhm... well:
(in any order)
-----
So, last year the original goal was simply to organize a blind test of midrange/fullrange drivers and find the best ones. Pretty straightforward.
Then, preliminary tests showed that it WON'T be that simple. Nope.
It seemed that, once blinded, it's much harder than we thought just to identify said drivers.
So i did what i had to do: push the organization of the whole thing on a more serious/scientific level. Took time and planifications, and then more adjustements and trials, etc... Bottomline, you can read the 100+ pages to understand the whole process if you want, but basically the conclusions (so far*) are the following:
1. Auditory capacities of humans are massively overestimated by audiophiles (and probably by most humans as well)
2. Frequency Response is King.
3. Once EQ'd, a 10$ midrange can mimic a 1500$ midrange, if within mechanical/electrical limits.
4. DSP/EQ/in-room measure tools might be the best investment an audiophile can make in our era.
5. Others will have to continue spending hundreds and thousands for a natural uncorrected FR.
*so far, because for the sake of statistical validity, i'll continue with the blind test on more participants.
-----------------
Here is the results of Matchup #1
World's best midrange Blind Testing - Need your help. - Page 115 - diyAudio
That is the easiest and most forgiving drivers matchup i could find. In fact, as you can see in the thread, i had to ''cheat'' with ½ octave difference... Expected success rate was very high.
(will be updated with most recent results)
As of 27th august 2016:
45 successful trials out of 61 (74% success rate)
3 successful participants out of 6 (50% success rate)
.
Where to start? Let's start by the beginning, which was 1 year ago when i started that thread:
World's best midrange Blind Testing - Need your help.
Now, 100+ pages and 65,000 views later i've come to some conclusions that are... uhm... well:
(in any order)-----
So, last year the original goal was simply to organize a blind test of midrange/fullrange drivers and find the best ones. Pretty straightforward.
Then, preliminary tests showed that it WON'T be that simple. Nope.
It seemed that, once blinded, it's much harder than we thought just to identify said drivers.
So i did what i had to do: push the organization of the whole thing on a more serious/scientific level. Took time and planifications, and then more adjustements and trials, etc... Bottomline, you can read the 100+ pages to understand the whole process if you want, but basically the conclusions (so far*) are the following:
1. Auditory capacities of humans are massively overestimated by audiophiles (and probably by most humans as well)
2. Frequency Response is King.
3. Once EQ'd, a 10$ midrange can mimic a 1500$ midrange, if within mechanical/electrical limits.
4. DSP/EQ/in-room measure tools might be the best investment an audiophile can make in our era.
5. Others will have to continue spending hundreds and thousands for a natural uncorrected FR.
*so far, because for the sake of statistical validity, i'll continue with the blind test on more participants.
-----------------
Here is the results of Matchup #1
World's best midrange Blind Testing - Need your help. - Page 115 - diyAudio
That is the easiest and most forgiving drivers matchup i could find. In fact, as you can see in the thread, i had to ''cheat'' with ½ octave difference... Expected success rate was very high.
(will be updated with most recent results)
As of 27th august 2016:
45 successful trials out of 61 (74% success rate)
3 successful participants out of 6 (50% success rate)
.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
But the famous BBC, no one have thinked to find a pleasant targett curve at ears then try to reproduce it on a speaker. (so the furthest normalisation curves we saw in the pro audio world...studio, cinema, etc)
With the recording Tools on smartphones, it could be pleasant people capture such curves on the famous speakers (the branded ones) and look for an existing trade off about a pleasantt targett curve...
Idem during live events in good concert room/hall !
Of course if the frequency response is king ?!
what we also known is that the pleasant frequency response curve is a combo with the room... even with of axis controlled directivity, there is also the bass: so it's certainly true than DSP helps...
So what is the good targett and in witch spl Windows (dynamic) ??? if we have to EQ....
For me the quality of the source helps a lot too, not sure a MiniDSP fights a good DAC, but if we have a roadmap to make an average acceptable sounds with a targett curve to copy: sure it is a big progress for most of the diyers to active EQ and mostly in relation to the Q/P (we spend a lot, not always for the best !)
Now can you simulate a QUAD ESL57 with a wide range paper driver ????
Can we really deal with the room and reflexion with a MiniDSP ? By playing with time delay/FR combo ????
Plus : I never saw several of the best tweeters having the same curve alones, if we speak now about the best trebles !
But it never hurts to make some workshops like Jon.
Has anyone readed this book from Carleen Maley Hutchins about violin and sound perception and reproduction (by "cloning" a good violin) ? "Research papers in violin acoustics 1975-1993" ; It was talked by Jean Hiraga in la Nouvelle revue du son issue 284, april 2004... we rasily understand than if frequency response would be the king, this last should not to be srtaight flat like the hifi magasines writed...or some writes yet !
With the recording Tools on smartphones, it could be pleasant people capture such curves on the famous speakers (the branded ones) and look for an existing trade off about a pleasantt targett curve...
Idem during live events in good concert room/hall !
Of course if the frequency response is king ?!
what we also known is that the pleasant frequency response curve is a combo with the room... even with of axis controlled directivity, there is also the bass: so it's certainly true than DSP helps...
So what is the good targett and in witch spl Windows (dynamic) ??? if we have to EQ....
For me the quality of the source helps a lot too, not sure a MiniDSP fights a good DAC, but if we have a roadmap to make an average acceptable sounds with a targett curve to copy: sure it is a big progress for most of the diyers to active EQ and mostly in relation to the Q/P (we spend a lot, not always for the best !)
Now can you simulate a QUAD ESL57 with a wide range paper driver ????
Can we really deal with the room and reflexion with a MiniDSP ? By playing with time delay/FR combo ????
Plus : I never saw several of the best tweeters having the same curve alones, if we speak now about the best trebles !
But it never hurts to make some workshops like Jon.
Has anyone readed this book from Carleen Maley Hutchins about violin and sound perception and reproduction (by "cloning" a good violin) ? "Research papers in violin acoustics 1975-1993" ; It was talked by Jean Hiraga in la Nouvelle revue du son issue 284, april 2004... we rasily understand than if frequency response would be the king, this last should not to be srtaight flat like the hifi magasines writed...or some writes yet !
Last edited:
It's taken me a while to get there, but I have to agree with the above. There will always be different or maybe even better ways to conduct these test but I have come to this conclusion in my own room with no abx, no time limits and very little monetary restraint. With a few exception our ears are easily fooled.
Flawed sure, but a waste of time, no. How about you set up a proper test and share your findings, any additional research is welcomed.
Well indeed the test was flawed, most are for one reason or another but as mordikai says, not a waste of time. No test is really a complete waste of time it's how you choose to interpret the data.
I am not surprised by the findings either. This is why I build to specifications rather than to what price a driver commands. Sometimes spending a boatload is necessary to give you what you need/want. Sometimes it is not.
I am not surprised by the findings either. This is why I build to specifications rather than to what price a driver commands. Sometimes spending a boatload is necessary to give you what you need/want. Sometimes it is not.
Here's one small example of why these sort of tests are a waste of time (I have plenty more examples).
Over 20 years ago I bought a pair of very expensive Dynaudio contour 1.3se speakers to replace my budget Kef Coda 8.
I bought the Dynaudios secondhand, so they were well run-in (if you believe speakers run-in - Some people don't). All the reviews I'd read had said they were very good speakers and I liked what I'd heard in the shop demo, so obviously I was expecting them to sound good (expectation bias).
It took a few weeks to realise that I was listening to music less and less and for some strange reason I wasn't enjoying music anymore.
I plugged the old Kefs back into the system and all the fun came back and I wanted to listen to my music again.
Both of my amplifiers sound very different in my system but I wouldn't be at all surprised if I couldn't tell a difference between my amps in someone else's system because I'd be unfamiliar with how the rest of it sounds.
Obviously I can hear some of the bigger differences with these blind ABX type tests but I think it's much easier to be fooled. I've found the best thing that works for me is to live with a whole system for a period of time (a couple of weeks would probably be enough) and then change one thing - Amp, speakers etc.
You're right. What I should have said is the test would be a waste of time for me.
Epic post #2, my dear friend. Epic.
I don't think you quite grasp the very meaning of my conclusions, so, let me put it this way:
I could take a pair of Sonus Faber Aida (120,000$+) and MIMIC the sonic result with... few cheap parts ordered from Parts-express, and an EQ.
Even better than better sounding, i can MIMIC the damn thing.... so 90% or 99.9% of people wouldnt even be able to make the difference.
No, my dear friend. I didnt waste time. I accidentally nuked Audiophilia as we know it.
.
.
Last edited:
Yes, I guess if you concluded that FR is the larger factor and it is often the main audible difference and sometimes the only one, then that's pretty significant. It's also what I suspected. Some might find it a jagged pill to swallow. If only there were more agreement so the results had more credibility, it would directly answer many of the questions here.
For the rest of us we can continue to talk about the baffle and the crossover and some of the more broad aspects of sensible driver choices as well as the others (like cone size) that are not the same as the kind of driver choices talked about here.
For the rest of us we can continue to talk about the baffle and the crossover and some of the more broad aspects of sensible driver choices as well as the others (like cone size) that are not the same as the kind of driver choices talked about here.
Ok guys, i admit. It's a HUGE pill to swallow. I myself (mr. exotic drivers) can see my whole audiophile's life passing through my head like a big joke.
But, anybody that wants to help/co-finance the thing to go even further, i'm in.
It's pretty easy, in fact: let's take real commercial ''exotic'' loudspeakers and let's prove my theory with large number or participants..
Name (and supply!) your champion: Sonus Faber, B&W, Wilson Audio...
Then, we contact Consumer reports, etc... with the results.
But, anybody that wants to help/co-finance the thing to go even further, i'm in.
It's pretty easy, in fact: let's take real commercial ''exotic'' loudspeakers and let's prove my theory with large number or participants..
Name (and supply!) your champion: Sonus Faber, B&W, Wilson Audio...
Then, we contact Consumer reports, etc... with the results.
It's old news. Earl Geddes already determined the relatively small contribution of the driver by equalising a few of them in similar tests to this one. He happened to apply it to waveguides, which are similar to baffles.. but easy to dismiss as different.. He also referred to drivers that didn't meet the basic requirements of the design as 'broken'. An expedient term, but useful nonetheless.
Don't forget that in this test, you equalized only the on-axis response. I know speaker(s) that only sound good on-axis. These are not mine. I have never listened to music in a single sweet spot...
In order to make your finding true, you need at least to design your speaker with a wave guide. Even then, no, I'm sure you cannot mimic a good speaker (usually expensive) with cheap drivers...
Like I said many times, my bottleneck is the non-linear distortion. And in your midrange test I want only to listen to (or look for) NLD and "details". And like I said before, I prefer seeing the measurement results and T/S parameters of the drivers than a conclusion of such test, in order to help me make decision. From this POV, Fatmarley could be right, for some people it could be a waste of time... (For others who don't know, this test could be revealing and useful)
Last edited:
My conclusions as well that frequency response is king, although not the only factor in sound. There exists other factors such as high resolution frequency response, which is difficult to measure, but can distinguish between soft and rigid cones. Anyone with a good EQ can convince themselves that EQ can make speakers sound alike.
That said, I don't like audio blind tests because they are very insensitive at detecting differences. They are meaningful if a difference is detected, but meaningless if no difference is found. A quite large difference in sound can exist but very difficult to distinguish with tests that are not sensitive. You cannot even distinguish 5% harmonic distortion yet it easily affects your emotional response to music over periods of time.
That said, I don't like audio blind tests because they are very insensitive at detecting differences. They are meaningful if a difference is detected, but meaningless if no difference is found. A quite large difference in sound can exist but very difficult to distinguish with tests that are not sensitive. You cannot even distinguish 5% harmonic distortion yet it easily affects your emotional response to music over periods of time.
So did you measure the speakers? All your story tells us is that you liked one speaker more than the other. If you measured them you might know why you preferred one over the other.
I trust my ears (and brain) and fantastic taste in music
to provide me with wonderful listening experiences.I trust measurements and science to explain why.
Last edited: